Data Visualization

• Open RStudio

• Help > Cheatsheets > Data Visualization with ggplot2

• type ‘?’ in R console with function, package or data name

• Add “R” to a search with a copy of an error message

• Many talented programmers who scan the web and answer issues

https://stackoverflow.com/

Hadley Wickham Yihui Xie

https://yihui.name/en/2017/08/so-gh-email/

https://rmarkdown.rstudio.com/

R has several systems for making graphs

• Base R

participants_data <- read.csv("participants_data.csv")
plot(participants_data$academic_parents)

Bar plot of number of observations of binary data related to academic parents

plot(participants_data$academic_parents, participants_data$days_to_email_response)

Boxplot of days to email response grouped by binary data related to academic parents

Use help '?' for function

?plot

Many libraries and functions for graphs in R…

• ggplot2 is one of the most elegant and most versatile.

• ggplot implements the grammar of graphics to describe and build graphs.

• Do more and do it faster by learning one system and applying it in many places.

• Learn more about ggplot2 in “The Layered Grammar of Graphics”

http://vita.had.co.nz/papers/layered-grammar.pdf

qplot: 'poor man's ggplot?

library(ggplot2)
qplot(days_to_email_response, letters_in_first_name, data = participants_data)

Scatterplot of days to email response as a function of the letters in your first name

Use help '?' for function

?qplot

Want to understand how all the pieces fit together? See the R for Data Science book: http://r4ds.had.co.nz/

Example from your data

qplot(days_to_email_response, letters_in_first_name, color=academic_parents, size=working_hours_per_day, data=participants_data)

Scatterplot of letters in your first name as a function of days to email response with colors representing binary data related to academic parents and working hours per day as bubble sizes.

Make more graphs

Example from Anderson's iris data set

qplot(Sepal.Length, Petal.Length, data=iris, color=Species, size=Petal.Width)

Scatterplot of iris petal length as a function of sepal length with colors representing iris species and petal width as bubble sizes.

Use help '?' for data

?iris

qplot accepts formula arguments such as log

plot1<-qplot(carat, price, data = diamonds)
plot2<-qplot(log(carat), log(price), data = diamonds)

Use help '?' for data

?diamonds

#Create a sample

dsmall <- diamonds[sample(nrow(diamonds), 100), ]

#Plot with different colours for color
qplot(carat, price, data = dsmall, colour = color)

#Plot with different shapes for cut 
qplot(carat, price, data = dsmall, shape = cut)

Different colors and shapes

Use help '?' for function

?sample

Help on topic 'sample' was found in the following packages:

  Package               Library
  dplyr                 /Users/macbook/Library/R/3.5/library
  base                  /Library/Frameworks/R.framework/Resources/library


Using the first match ...

Set parameters manually with I()

qplot(carat, price, data = diamonds, alpha=I(0.1), colour=I("blue"))
qplot(carat, price, data = diamonds, alpha=I(0.4), colour=I("green"))

Inhibit Interpretation / Conversion of Objects

Use help '?' for function

?I

With “geom” different types of plots can be defined e.g. points, line, boxplot, path, smooth. These can also be combined in a vector.

qplot(carat,price,data=dsmall, geom="line")
qplot(carat,price,data=dsmall, geom="smooth")
qplot(carat,price,data=dsmall, geom=c("point","smooth"))

ggplot2 geom options

Use help '?' for function

?qplot

read 'Arguments' section of help file

Depending on your dataset size the smooth function will select different lines and smoothing methods.

qplot(carat,price,data=dsmall,geom=c("point","smooth"))
qplot(carat,price,data=diamonds,geom=c("point","smooth"))

With span the wiggliness of the line is controlled.

qplot(carat,price,data=dsmall, geom=c("point","smooth"), span=0.2)

Use method to specify your smoothing method

qplot(carat,price,data=dsmall,geom=c("point","smooth"),method="lm")

ggplot2 lines and smoothing options

• Boxplots can be displayed through geom=“boxplot”.

qplot(color,price/carat,data=diamonds,  geom="boxplot")

• Jittered plots (geom=“jitter”) show all points. In case of overplotting changing alpha can help.

qplot(color,price/carat,data=diamonds, geom="jitter")
qplot(color,price/carat,data=diamonds, geom="jitter", alpha=I(0.1))

ggplot2 boxplots and jitter

Histograms can be displayed through geom=“histogram”.

qplot(carat, data = diamonds, geom = "density")
qplot(carat, data = diamonds, geom = "density", colour = color)
qplot(carat, data = diamonds, geom = "density", fill = color, alpha=I(0.3))

ggplot2 histograms

Use factor to subset your data.

qplot(displ, hwy, data = mpg, colour = cyl, geom=c("point","smooth"),method="lm")

qplot(displ, hwy, data = mpg, colour = factor(cyl), geom=c("point","smooth"),method="lm")

ggplot2 subset with smooth line

• pulling out aes() from the ggplot() function:
  
• using fewer functions; example - using labs() to add a title instead of ggtitle()
• using functions multiple times; example aes(x = var1) + aes(y = var2) rather than aes(x = var1, y = var2)
• using base R functions and tidyverse functions. For other packages, the :: style to call them
• write out arguments (no shortcuts) aes(x = gdppercap) not aes(gdppercap)
• order ggplot commands so that reactivity is obvious; scale adjustments to aesthetics might also be near the aesthetic declaration.

https://evamaerey.github.io/ggplot_flipbook/ggplot_flipbook_xaringan.html#1

Usual ggplot code

ggplot(mtcars, aes(mpg, y = hp, col = gear)) +
  geom_point() +
  ggtitle("My Title") +
  labs(x = "the x label", y = "the y label", col = "legend title")

'Slow ggplotting' version for same plot

  ggplot(data = mtcars) +
  aes(x = mpg) +
  labs(x = "the x label") +
  aes(y = hp) +
  labs(y = "the y label") +
  geom_point() +
  aes(col = gear) +
  labs(col = "legend title") +
  labs(title = "My Title")

https://evamaerey.github.io/ggplot_flipbook/ggplot_flipbook_xaringan.html#1

cor.test(participants_data$days_to_email_response, participants_data$letters_in_first_name)

    Pearson's product-moment correlation

data:  participants_data$days_to_email_response and participants_data$letters_in_first_name
t = -1.647, df = 13, p-value = 0.1235
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
 -0.7649469  0.1229287
sample estimates:
       cor 
-0.4154989 

Use help '?' for function

?cor.test

• Using the datasauRus, ggplot2 and gganimate libraries.
• Note the many calls for ggplot

ggplot(datasaurus_dozen, aes(x=x, y=y))+
  geom_point()+
  theme_minimal() +
  transition_states(dataset, 3, 1) + 
  ease_aes('cubic-in-out')

• Using the gifski, ggplot2 and gganimate libraries.

ggplot(mtcars, aes(factor(cyl), mpg)) + 
  geom_boxplot() + 
  geom_point() +
  transition_states(am, transition_length = 4, state_length = 1) + 
  view_follow()

• Using the dplyr, ggplot2 and reshape2 libraries.

part_data<-select(participants_data, days_to_email_response, number_of_siblings, years_of_study, number_of_publications, letters_in_first_name, km_home_to_zef, working_hours_per_day, days_to_email_response)

cormat <- round(cor(part_data), 1)
melted_cormat <- melt(cormat)
ggplot(data = melted_cormat, aes(x=Var1, 
y=Var2, fill=value)) + 
geom_tile()

• Check with journal about size, resolution etc.

?pdf

?png

png(file = "cortile.png", width = 7, height = 6, units = "in", res = 300)

ggplot(data = melted_cormat, aes(x = Var1, y = Var2, fill = value)) + geom_tile() + theme(axis.text.x = element_text(angle = 45, hjust = 1))

dev.off()

list.files()

If time create and export more figures

• Check your data for interesting trends and correlations
• Use scatter plots, barcharts and boxplots
• Bootstrap and vary the sample and run the same analysis and plots
• Save your most interesting figure and share it with us tomorrow

• Import data from an external source (e.g. FAO, World Bank)
• Display those data in an interactive plot
• Play around with the design
• Export your most interesting figure and share it with us tomorrow

Install Git & Github (if you do not already have them).

Git https://git-scm.com/downloads

Github http://r-pkgs.had.co.nz/git.html

join Github https://github.com/